Nutritional Assessment: Predictive Variables at Hospital Admission Related with Length of Stay

Background: Studies indicate that 40-50% of hospitalized patients show malnutrition, a variable that is associated with length of stay and morbidity. The aim of our study was to detect nutritional parameters, which could have an influence on length of stay in hospitalized patients. Material and Methods: All patients with a nutritional evaluation at hospital admission were elegible for inclusion. A total of 1,088 patients were studied from January 1999 to December 2003. Length of stay (LOS) data was obtained from the patient hospital record after the patient was discharged. All patients received instruction in 24-hour written food record keeping. Albumin, prealbumin, transferrin, glucose levels and total lymphocytes, were measured in all patients. Weight, body mass index, tricipital skinfold, midarm muscle circumference and midarm muscle area were assessed in a standard way. Weight loss in the previous 3 months was recorded. Results: A total of 1,088 patients were enrolled, mean age 61.8 ± 17 years, weight 64.2 ± 15 kg and BMI 23.9 ± 4.6, with a weight loss 4.15 ± 9.6 kg. The sex distribution of patients was 65.2% male and 34.8% females. Distribution of diagnosis showed leukemia and lymphoma (11.5%), solid cancer (37.4%), infections (3.5%), neurological disease (13.6%), respiratory tract disease (8.8%), and miscellaneous group (25.3%). Length of stay was 29.45 ± 25.13 days. In whole group, the correlation analysis among length of stay (days) and predictive parameters showed a positive association between albumin and length of stay (r = -0.2; p < 0.05). In the multivariant analysis with a dependent variable (length of stay (days)) and independent variables with an association in univariant analysis adjusted by age and sex, only albumin remained as an independent predictor in the model (F = 8.8; p < 0.05), with an increase of 6.2 days (95% CI: 3.5-8.9) with each decrease of 1 g/dl of albumin. Conclusion: The serum albumin levels are a good marker of LOS, a decrease in admission levels produces an increase in LOS.